Enjoy listening to our podcast
Recent episodes
The SynapticCity Phenomenon: When All Foundation Models Marry Federated Learning and Blockchain
Authors: Roberto Gómez-Espinosa, Sergio Zaera
Abstract
Our work proposes an innovative framework for smart cities that integrates Foun- dation Models (FM), Federated Learning (FL) and Blockchain to address key challenges in urban data management, such as privacy, scalability and predictive accuracy. Combining the predictive power of FMs with the privacy-preserving capabilities of FL and the secure and transparent governance provided by Blockchain, we created a robust and decentralized solution to manage diverse urban data.
Our approach enables real-time data analysis and decision making, while ensuring that sensitive information remains secure. To demonstrate the effectiveness of this integrated platform, we present a use case in inventory management and sales forecasting for smart city companies, showing its potential to improve operational efficiency, data privacy and economic resilience. This synergy of advanced technologies sets a new standard for secure, adaptive and collaborative management.
Keywords: Smart Cities, Foundation Models, Federated Learning, Blockchain, Smart Contracts.
TL;DR: This paper proposes a framework for smart cities combining foundation models, federated learning, and blockchain to enhance data privacy, scalability, and predictive accuracy in urban environments.
Adaptive Federated Learning based on Blockchain: New Frontiers in Logistics and Stockpile Management
Authors: Zaera Mata, Sergio, Jimeno Paloma, Castañeda Pablo, Dorado Ander, Franco Jesús, Gómez-Espinosa Roberto
Abstract
Nowadays, defense and security operations require innovative and effective approaches for logistics and stockpile management. This paper proposes an ecosystem based on Federated Learning (FL), integrated with Blockchain (BC) and Smart Contracts (SC) technology, aimed at optimizing operational and strategic efficiency in the defense sector. The suggested platform capitalizes on advances in Internet of Things (IoT) for decentralized training of Artificial Intelligence (AI) models, which offers substantial improvements in data privacy, transactional security and resource management reliability. Through the synergistic application of FL, a collaborative strategy and autonomous model training that preserves privacy while enhancing the security and integrity of critical information management is ensured.
Blockchain provides an immutable structure for model registration, as well as a transparent and decentralized auditing mechanism, while the applied SCs reinforce the automation of operational processes and regulatory relationships. The feasibility and effectiveness of our system were demonstrated through the GREEN Project, where the implementation of these technologies enabled the optimization of electric vehicle (EV) charging stations as a successful case in the urban environment. Extending this methodology to the Defense context, the Logistics and Stockpile Management in the Security Sector (GLASS) concept is presented as a transformative approach to inventory management, offering improved logistics management, accurate resource demand determination and efficient alerting within highly regulated security environments.
Keywords: Blockchain, Digital Twin, Federated Learning, Inteligencia artificial, Physics-Informed Neural Networks, Smart Contracts.
Blockchain-based federated learning for logistics and stockpile management.
Authors: Zaera Mata, Sergio, Gallego Adrián, Pablo , Jimeno Sánchez-Patón, Paloma , Castañeda Fuentes, Pablo , Franco Moreu, Jesús , Gómez-Espinosa Martín, Roberto
Abstract
The revolution experienced by Artificial Intelligence (AI) techniques in the last decade deserves special attention. However, the conventional approach of centralized training of AI models poses substantial limitations and raises significant privacy concerns. Against this backdrop, it is imperative to seek alternatives that can leverage available information without compromising privacy, in order to generate mutual benefits for all parties involved. This study addresses the technical and collaborative challenges inherent to the implementation of AI technologies and proposes innovative solutions based on the Internet of Things (IoT).
Three primary challenges are identified: the limited computational capacity for massive volumes of data, the security and privacy imperatives set by regulations, and the reluctance of individuals and organizations to share sensitive information. To address these challenges, the adoption of the Federated Learning (FL) approach, combined with Blockchain (BC) and Smart Contracts (SC) technologies, is proposed.
To demonstrate the capabilities of this ecosystem, a use case is proposed in logistics and stockpile management in the Defense and Security sector, addressing challenges related to information security, reliability and traceability. This combination of technologies has been previously studied and applied in projects carried out by HI-Iberia, specifically in the GREEN project, which focuses on the optimization of electric vehicle charging stations, developed in collaboration with Naturgy, an expert company in the energy field.
Keywords: Inteligencia artificial, Federated Learning, Blockchain, Smart Contracts, Internet of things.
GREEN: Collaborative Intelligence for Sustainable Cities.
Authors: Zaera Mata, Sergio, Jimeno Paloma, Castañeda Pablo, Dorado Ander, Franco Jesús, Gómez-Espinosa Roberto
Abstract
The GREEN project focuses on developing collaborative intelligence for sustainable cities, with the goal of integrating secure and efficient artificial intelligence (AI) applications with Internet of Things (IoT) big data. The main use case is the optimization of electric vehicle (EV) charging stations to predict energy demand, negotiate prices efficiently and improve infrastructure utilization. The project employs Federated Learning (FL) for privacy-preserving predictive models, Blockchain for decentralized and secure data management, and Smart Contracts (SC) to automate processes and ensure transparency. Through these technologies, GREEN facilitates distributed AI model training while maintaining data locality, improving efficiency in urban resource management, and reducing energy waste.Translated with DeepL.com (free version)
The platform not only focuses on EV charging, but also aligns with broader Smart City initiatives, with potential applications in public transportation, environmental monitoring and urban planning. Key outcomes include significant reductions in greenhouse gas emissions, improved air quality and lower urban noise levels. The project supports the integration of renewable energy into charging stations and promotes energy efficiency and equitable access, particularly in rural areas. GREEN's holistic approach addresses urban challenges while fostering sustainable development and technological innovation, contributing to a future of resilient, inclusive and environmentally friendly cities.
Keywords: Artificial Intelligence, Federated Learning, Internet of Things, Blockchain, Smart Contracts, Smart Cities, Sustainability, Charging Stations, Energy Optimization, Urban Mobility, Electric Vehicles, Resource Management, Data Privacy, Transparency, Emission Reduction.